An Overview of Plume Tracker: Mapping Volcanic Emissions with Interactive Radiative Transfer Modeling

Monday, 15 December 2014
Vincent J Realmuto1, Alexander Berk2 and Chona Guiang2, (1)Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, United States, (2)Spectral Sciences Inc., Burlington, MA, United States
Infrared remote sensing is a vital tool for the study of volcanic plumes, and radiative transfer (RT) modeling is required to derive quantitative estimation of the sulfur dioxide (SO2), sulfate aerosol (SO4), and silicate ash (pulverized rock) content of these plumes. In the thermal infrared, we must account for the temperature, emissivity, and elevation of the surface beneath the plume, plume altitude and thickness, and local atmospheric temperature and humidity. Our knowledge of these parameters is never perfect, and interactive mapping allows us to evaluate the impact of these uncertainties on our estimates of plume composition.

To enable interactive mapping, the Jet Propulsion Laboratory is collaborating with Spectral Sciences, Inc., (SSI) to develop the Plume Tracker toolkit. This project is funded by a NASA AIST Program Grant (AIST-11-0053) to SSI. Plume Tracker integrates (1) retrieval procedures for surface temperature and emissivity, SO2, NH3, or CH4 column abundance, and scaling factors for H2O vapor and O3 profiles, (2) a RT modeling engine based on MODTRAN, and (3) interactive visualization and analysis utilities under a single graphics user interface.

The principal obstacle to interactive mapping is the computational overhead of the RT modeling engine. Under AIST-11-0053 we have achieved a 300-fold increase in the performance of the retrieval procedures through the use of indexed caches of model spectra, optimization of the minimization procedures, and scaling of the effects of surface temperature and emissivity on model radiance spectra. In the final year of AIST-11-0053 we will implement parallel processing to exploit multi-core CPUs and cluster computing, and optimize the RT engine to eliminate redundant calculations when iterating over a range of gas concentrations. These enhancements will result in an additional 8 – 12X increase in performance. In addition to the improvements in performance, we have improved the accuracy of the Plume Tracker retrievals through refinements in the description of surface emissivity and use of vector projection to define the misfit between model and observed spectra.

Portions of this research were conducted at the Jet Propulsion Laboratory, California Institute of Technology, under contract to the National Aeronautics and Space Administration.